Read on Twitter
The notion of “Ethical AI” has been leveraged by #bigtech to push for voluntary "AI Principles" over government regulation.
But there are huge gaps in their programs that have very harmful, real-life effects.
An equitable and accountable AI thread.
But there are huge gaps in their programs that have very harmful, real-life effects.
An equitable and accountable AI thread.
Using ethics in AI is not itself problematic but terms like "Ethical AI" and "AI Principles" have been co-opted.
Right now, "Ethical AI" allows companies and govs to follow rules they've set for themselves.
The ball's in their court from start to finish. https://www.bostonglobe.com/2021/04/27/business/google-pressured-racial-equity-audit-after-collapse-its-ai-ethics-team/
Right now, "Ethical AI" allows companies and govs to follow rules they've set for themselves.
The ball's in their court from start to finish. https://www.bostonglobe.com/2021/04/27/business/google-pressured-racial-equity-audit-after-collapse-its-ai-ethics-team/
The self-regulating "Ethical AI" approach is very limited because it doesn't create any mandatory requirements or ban certain uses of AI.
Both words AND actions matter.
We need to bridge the gap from principles to practice with #EquitableAI and #AccountableAI.
Both words AND actions matter.
We need to bridge the gap from principles to practice with #EquitableAI and #AccountableAI.
#EquitableAI works to:
Give people a voice and a choice in how they interact with AI
Get consent first
Respect human rights and dignity
Prevent AI from being used by the powerful to increase their control
Require prohibiting certain government and for-profit uses of AI
Give people a voice and a choice in how they interact with AIGet consent firstRespect human rights and dignityPrevent AI from being used by the powerful to increase their controlRequire prohibiting certain government and for-profit uses of AI
#AccountableAI works to:
show how the AI system works and its design
be transparent and clear about its specific purpose
require continuous oversight by 3rd parties
give people a way to contest and address any harm they've experienced
show how the AI system works and its design be transparent and clear about its specific purpose require continuous oversight by 3rd parties give people a way to contest and address any harm they've experienced
Real #AlgorithmicJustice means the wealthy can't use AI to tighten their control over people — especially where it adds to systematic oppression like racial profiling in law enforcement, gender bias in hiring, and over-policing of migrant communities. https://www.cnn.com/2021/04/29/tech/nijeer-parks-facial-recognition-police-arrest/index.html
There are many terms used to describe a better approach to AI out there.
But in order to have truly just AI systems, equity and accountability need to be baked in from the start and revisited continuously.
Join the fight http://bit.ly/ajlnewsletter-signup
#EquitableAI #AccountableAI
But in order to have truly just AI systems, equity and accountability need to be baked in from the start and revisited continuously.
Join the fight
http://bit.ly/ajlnewsletter-signup#EquitableAI #AccountableAI
